About GenPeach AI

GenPeach AI is a product-driven research lab aiming to redefine how people create and interact through multimodal, emotionally resonant AI.

We are building vertical foundation models specializing in generating hyper-realistic humans in image & video. Our stack involves working with PB-scale proprietary datasets, designing novel model architectures, efficiently training them on large GPU clusters and integrating them in the end-user products.

We train and deploy our own large-scale models and ship them into real products. Our team operates at the intersection of research-grade AI and production-grade systems engineering.

About the Role

We are looking for a Member of Technical Staff (MTS) to own and evolve the Large Scale Data Infrastructure layer that powers both research and production at GenPeach AI.

This is a data infrastructure-first role: your primary mission is to design and operate the systems that store, process, and serve petabytes of image and video data — and to run the large-scale GPU inference pipelines that annotate, transform, and enrich that data at scale. The datasets you build and maintain will directly fuel the training of our foundation models.

This is a high-ownership, high-impact role at the core of what makes GenPeach's research and products possible.

In this role, you will

• Build and maintain systems that ingest raw image, video, and multimodal data from many sources and turn it into high-quality, training-ready datasets at petabyte scale.

• Design and operate large-scale data pipelines for processing, annotation and dataset generation.

• Build and maintain infrastructure for large-scale dataset analytics to support dataset inspection, curation, and quality analysis.

• Optimize distributed inference workloads used for data annotation across many GPUs, with strong focus on throughput, efficiency, and reliability.

• Build efficient storage, retrieval, and migration systems for petabyte-scale datasets across object storage, clouds, and infrastructure providers.

• Ensure training datasets can be consumed efficiently in any cloud or provider environment, with data loading and storage throughput not becoming bottlenecks during distributed training.

• Improve training infrastructure reliability, monitoring, debuggability, and utilization on large GPU clusters.

• Build and maintain infrastructure-as-code, observability, and operational tooling across data and training workloads.

• Help build scalable, provider-agnostic infrastructure for production GPU inference services powering our creative platform.

• Work closely with ML, research, backend, and data teams, and contribute to long-term infrastructure decisions.

Minimum Qualifications

• 5+ years of professional software engineering experience (Python)

• Hands-on experience managing large-scale datasets (hundreds of TBs or more) on object storage (S3 or equivalent)

• Experience building and operating large-scale GPU inference workloads: job queues (e.g. SQS, Kafka, RabbitMQ) dispatching AI model inference jobs across large fleets of GPUs

• Kubernetes at scale: orchestrating large numbers of pods for data-processing or inference workloads

• Strong Python proficiency, including async programming, concurrency/multiprocessing, and performance optimization

• Experience operating production systems in Linux environments

• Hands-on experience with Docker and Infrastructure-as-Code (Terraform or similar)

Preferred Qualifications

• Experience specifically with large-scale image or video datasets (PB-scale)

• Experience with workflow orchestrators: Ray, Dagster, Airflow, Slurm, or similar

• Familiarity with model serving architectures and inference optimization

• Exposure to observability stacks (metrics, logs, tracing) for ML systems

• Experience with distributed model training (multi-GPU / multi-node) — purely a bonus, not a requirement

• Strong fundamentals in data structures and algorithms

What Makes This Role Unique

• You own the data layer that directly determines what our foundation models can learn — model quality starts with you

• You'll run GPU inference at a scale most engineers never touch: hundreds of GPUs processing petabytes of image and video

• Research and infra work side by side — your decisions quickly show up in model results

• Small senior team, high trust, no bureaucracy

Our Culture

• High ownership and accountability

• Strong technical standards

• Direct, low-ego communication

• Bias toward shipping, measuring, and iterating fast
  

Logistics

• Location: Zurich or Warsaw: onsite or hybrid. If you’re elsewhere, we’re open to remote (team/timezone fit considered).

• Competitive salary + meaningful equity (depending on role and level)

• Interview process: quick screen → technical (practical + systems) → team fit/values